Yahoo Groups archive

AVR-Chat

Index last updated: 2026-04-28 22:41 UTC

Thread

proper power supply filtering

proper power supply filtering

2005-05-16 by arhodes19044

I am trying to design a PCB for an ATmega128.  I probably will use 
an Olimex header board on top my my PCB.

I want to supply good 5v to the header board and the rest of my 
project.  I plan on using some LM7805 type of device, but since it 
will be supplying everything, including the back lights of the LCD 
displays which are surprisingly rated at several hundred milliamps 
EACH, the supply will have to be rated at 1.5 amps to be sure that 
everything is covered plus insurance.

I will eventually be drawing 13.x volts from a potentially fairly 
noisy car power source.  The supply needs reasonably good 
filtering.  I want good filtering without overkill.

What filtering is necessary?  I assume a pretty big electrolytic on 
the 13v side ( and maybe a tantalum or something?), and something 
else on the other side.  On the output side I have seen 2 caps on 
some schematics, but the sizes and types were not listed.

I am sure this is a problem which is well known and solved for this 
application, so it might be easy to answer.  What specific types and 
sizes of caps are usually recommended?

-Tony

Re: proper power supply filtering

2005-05-16 by arhodes19044

--- In AVR-Chat@yahoogroups.com, Dave VanHorn <dvanhorn@d...> wrote:
> 
> Avoid tantalums.
> 
> You have bigger problems.
> Have you checked the max current rating and max heat dissipation 
> ratings for the regulator?
> 
> Also, the LM78XX series is not designed to run from vehicle power.
> National makes a few that are though, and it won't be a painful 
change.
> 
> 
> 0.1uF on input and output, located as close to the reg as 
possible, 
> to stop oscillation.
> You will need to add electrolytics on the input and output as 
> required by the circuit, but they do not take the place of the 0.1 
caps.
> You also need a reverse-biased diode across the regulator's input 
and 
> output, to prevent destruction of the reg, if there is a short on 
the 
> input side.

Thanks for the quick reply!

I am a definite beginner as far as design goes.  Why not tantalums?  
I suppose that the alternative would be ceramic?

I had not checked the maximum input voltage on the 7805, Is it less 
than 13.6v?  Or is there some other automotive power issue at work 
here?

Also, another sign that I know nothing, the reverse bias diode 
across the input and output of the reg.  I suppose that if the input 
goes to zero while there is +5 on the output this will cause the 
damage?  How big a diode do I need.  A 1A rectifier rather than 
a "switching" diode, not that I know the difference except in 
current ratings and possibly switching times....  The design apps I 
have seen did not show such a diode, but it seems to make sense!

I ask these questions not to indicate that I do not believe the 
answers, just the opposite.  I need to learn the reasons so I will 
not make the same sorts of mistakes in the future!!!!

As for the large electrolytics that the design may require.  Would I 
need one on BOTH the input and output?  I would have thought that 
one on the input would be good enough, but I can see that big 
transients may not get fed well through the reg, so a big cap on the 
output would be beneficial too....

-Tony

Re: [AVR-Chat] proper power supply filtering

2005-05-16 by Dave VanHorn

>
>What filtering is necessary?  I assume a pretty big electrolytic on
>the 13v side ( and maybe a tantalum or something?), and something
>else on the other side.  On the output side I have seen 2 caps on
>some schematics, but the sizes and types were not listed.

Avoid tantalums.

You have bigger problems.
Have you checked the max current rating and max heat dissipation 
ratings for the regulator?

Also, the LM78XX series is not designed to run from vehicle power.
National makes a few that are though, and it won't be a painful change.

>I am sure this is a problem which is well known and solved for this
>application, so it might be easy to answer.  What specific types and
>sizes of caps are usually recommended?

0.1uF on input and output, located as close to the reg as possible, 
to stop oscillation.
You will need to add electrolytics on the input and output as 
required by the circuit, but they do not take the place of the 0.1 caps.
You also need a reverse-biased diode across the regulator's input and 
output, to prevent destruction of the reg, if there is a short on the 
input side.

Re: [AVR-Chat] Re: proper power supply filtering

2005-05-16 by Mike Perks

Tony,

I don't know all the answers to your questions but here are a few:

> --- In AVR-Chat@yahoogroups.com, Dave VanHorn <dvanhorn@d...> wrote:
>
> I had not checked the maximum input voltage on the 7805, Is it less
> than 13.6v?  Or is there some other automotive power issue at work
> here?

Maximum is well above 13.6V. However the larger the difference, the more 
heat needs to be disappiated by the regulator. You also mentioned that 
you will need 1.5A which exceeds the specifications of the LM7805 which 
is limited to 1A. You will need something like a MC78T05 that is rated 
at 3A.

> Also, another sign that I know nothing, the reverse bias diode
> across the input and output of the reg.  I suppose that if the input
> goes to zero while there is +5 on the output this will cause the
> damage?  How big a diode do I need.  A 1A rectifier rather than
> a "switching" diode, not that I know the difference except in
> current ratings and possibly switching times....  The design apps I
> have seen did not show such a diode, but it seems to make sense!

I use 1N4001 or 1N4002 to protect both against polarity reversal and to 
protect the regulator against reverse current flow (connect the cathode 
to the regulator input). For your case you are going to need beefier 
diodes such as 1N5402.

Regards,
Mike

Re: [AVR-Chat] Re: proper power supply filtering

2005-05-16 by Dave VanHorn

>
>I am a definite beginner as far as design goes.  Why not tantalums?

Tantalums have very few advantages, and a lot of problems.
They like to explode when hit with too much current, and that can 
easily happen with vehicle power.

>I suppose that the alternative would be ceramic?
>
>I had not checked the maximum input voltage on the 7805, Is it less
>than 13.6v?  Or is there some other automotive power issue at work
>here?

400V load dump transients,  inductive spikes to 30+ volts.

>Also, another sign that I know nothing, the reverse bias diode
>across the input and output of the reg.  I suppose that if the input
>goes to zero while there is +5 on the output this will cause the
>damage?  How big a diode do I need.  A 1A rectifier rather than
>a "switching" diode, not that I know the difference except in
>current ratings and possibly switching times....  The design apps I
>have seen did not show such a diode, but it seems to make sense!

Few app notes show it these days, but it is required by almost all of 
these sorts of regulators.


>As for the large electrolytics that the design may require.  Would I
>need one on BOTH the input and output?  I would have thought that
>one on the input would be good enough, but I can see that big
>transients may not get fed well through the reg, so a big cap on the
>output would be beneficial too....

Exactly.

I would recommend a diode in series with the input, to protect from 
negative voltage transients, and also to help with your power 
dissipation problem. A series fuse at 3A, and an 20V 1W zener, 
reverse biased across the input, behind the fuse, should be enough to 
keep your project alive.  An electrolytic inside this, in front of 
the regulator, will help you survive transients on the supply line.

You might consider splitting the supplies, and running your logic 
from one, and the high current loads from the other.
You could then use a large enough electrolytic to power your logic 
for several seconds during cranking, and let the high current devices glitch.
You'd have to deal with any consequences of that, but it would also 
help spread the load off that very very warm regulator.

At 1.5A you'll need a 3A diode, and it will dissipate about 1W. 
Meanwhile, your regulator will be dissipating about 14W, assuming it 
is rated for 1.5A output, and it will require a large heat sink. 
Worse, the ambient temperature in a car can easily pass 140F, which 
makes the heatsink requirement worse.  Pd = (Vin - Vout) * Iout  so 
(14.4 - 5) * 1.5 = 14.1W   I didn't factor in the diode drop, but I 
also didn't worst-case it at 15V, which pretty much balances 
out.   I've used 15W soldering irons, that had more heat sink area 
than that regulator, if that helps put those numbers into perspective.

A switching regulator would make these heat problems pretty much disappear.
Have a look at the design guides on the national semiconductor site 
for their simple switchers.
You'll be looking at a "buck" topology.

Re: proper power supply filtering

2005-05-16 by alwelch93021

1. Be sure to read  the data sheet for the regulators.  I just 
redesigned my board and chose to use a national semi LDO which does 
not like MLC caps.  I happen to also like MLC caps as generally they 
work great.

I also had a heavy load which i found drooped my input source enough 
that my switching regulator would also droop 20 millivots. So what i 
did was put the heavy load on  the 12VDC incomming supply and I 
used  the LDO regulator to regulate that 12VDC down to 10.2 and  
then on to my switcher which generatest the +/-5 and 3.2VDC.  Now my 
analog and AVR will not see the droop.

I also recommend ferrite beads on  the incomming power and ground 
lines.  I found this helps with EMC filtering.

Al Welch

--- In AVR-Chat@yahoogroups.com, "Mike Murphree" <mike@t...> wrote:
> Jim Wagner said:
> >
> > Tantalum caps are nice because of small size and low ESR.
> > ESR isn't much of a requirement in linear regulators. But,
> > they cost more, have limited voltages and values, and have
> > a tendancy to destroy themselves (ie, explode or burst into
> > flames) if not in a current-limited circuit. Thus, putting
> > them in a place where they directly "see" battery power is
> > asking for trouble. It is less of an issue on the output
> > since the regulator, itself, provides current limiting.
> > They also tend to be more attractive on the output side
> > since the larger value is needed there. A general rule
> > commonly used by designers is to specify a tantalum voltage
> > rating that is about 2X the actual working voltage.
> 
> We have typically used MLC (multilayer ceramics) on switching 
supplies
> with some additional electrolytics if needed for bulk 
capacitance.  The
> MLC capacitors have an extremely low ESR and can be more effective 
than
> tantalum capacitors.  Also ceramic capacitors are more tolerant of 
applied
> voltages and are often tested to 4X or more of rated value at the 
factory.
> 
> Mike

Re: proper power supply filtering

2005-05-16 by arhodes19044

--- In AVR-Chat@yahoogroups.com, Mike Perks <basicx@a...> wrote:
> Maximum is well above 13.6V. However the larger the difference, 
the more 
> heat needs to be disappiated by the regulator. You also mentioned 
that 
> you will need 1.5A which exceeds the specifications of the LM7805 
which 
> is limited to 1A. You will need something like a MC78T05 that is 
rated 
> at 3A.
> 
> I use 1N4001 or 1N4002 to protect both against polarity reversal 
and to 
> protect the regulator against reverse current flow (connect the 
cathode 
> to the regulator input). For your case you are going to need 
beefier 
> diodes such as 1N5402.
> 
> Regards,
> Mike

Thanks Mike.  I said that I need 1.5A mainly because I am using a 
1.5A reg right now and it only gets moderately warm, but not hot 
with the loads right now.  There would be a second LCD which is the 
major single load.  I believe that a 1A reg would work, but as I 
said, I'd rather over-build the supply section.

I will use some big diodes as protection.  Already I use a diode on 
the input to protect against reverse polarity 13.6V input.  This 
will drop the input voltage by a volt or so right there, taking some 
of the thermal load off the regulator, I suppose.

-Tony

Re: proper power supply filtering

2005-05-16 by arhodes19044

--- In AVR-Chat@yahoogroups.com, Dave VanHorn <dvanhorn@d...> wrote:

> Have a look at the design guides on the national semiconductor site 
> for their simple switchers.
> You'll be looking at a "buck" topology.

I saw those devices just today in the catalog.  I was wondering if 
they are any good.  If so, I can probably make good use of one of them!

-Tony

Re: proper power supply filtering

2005-05-16 by arhodes19044

--- In AVR-Chat@yahoogroups.com, "alwelch93021" <alwelch@a...> wrote:
> 
> I also recommend ferrite beads on  the incomming power and ground 
> lines.  I found this helps with EMC filtering.
> 

good idea!

On a similar unit (one that I am reverse-engineering), I ran it on 
an old (pre-electronics) car with copper wire ignition wires.  The 
electronics worked perfectly with the engine off, but went INSANE 
with the engine running.  I had to put one of those inductive 
filters (the ones that look like a small transformer) in the power 
line to the electronics.

It still glitched majorly when a nearby big fat relay and its 
solenoid would fire.  I had to make sure never to use THAT circuit 
when running the electronics.  I might have been able to put a diode 
and cap across the relay, but I was not sure if it would work and I 
never got to it.


-Tony

Re: [AVR-Chat] Re: proper power supply filtering

2005-05-16 by Jim Wagner

On the reverse diode between the input and output of the
regulator: this is normally used to prevent reverse current
flow through the regulator. This could occur if the input
is suddenly forced to a low voltage while the output cap is
still charged. The LM78xx family is not as sensitive to
this as some and I rarely see this diode in 7805 circuits. 

Also, don't forget power dissipation! If your load current
is 1A, and the input voltage is 13.5V, the voltage drop is
8.5V. The 1A and 8.5V makes 8.5W. This is a lot of power
and requires some careful thinking and design to make sure
the head is removed and the junction temperature is not too
high. If It does get too warm, it will shut down, often
leading to mysterious behaviour of the circuit.

Tantalum caps are nice because of small size and low ESR.
ESR isn't much of a requirement in linear regulators. But,
they cost more, have limited voltages and values, and have
a tendancy to destroy themselves (ie, explode or burst into
flames) if not in a current-limited circuit. Thus, putting
them in a place where they directly "see" battery power is
asking for trouble. It is less of an issue on the output
since the regulator, itself, provides current limiting.
They also tend to be more attractive on the output side
since the larger value is needed there. A general rule
commonly used by designers is to specify a tantalum voltage
rating that is about 2X the actual working voltage. 

Jim

---------------------------------------------------------------
The Think Different Store
http://www.thinkdifferentstore.com/
For All Your Mac Gear
---------------------------------------------------------------

Re: [AVR-Chat] Re: proper power supply filtering

2005-05-16 by Mike Murphree

Jim Wagner said:
>
> Tantalum caps are nice because of small size and low ESR.
> ESR isn't much of a requirement in linear regulators. But,
> they cost more, have limited voltages and values, and have
> a tendancy to destroy themselves (ie, explode or burst into
> flames) if not in a current-limited circuit. Thus, putting
> them in a place where they directly "see" battery power is
> asking for trouble. It is less of an issue on the output
> since the regulator, itself, provides current limiting.
> They also tend to be more attractive on the output side
> since the larger value is needed there. A general rule
> commonly used by designers is to specify a tantalum voltage
> rating that is about 2X the actual working voltage.

We have typically used MLC (multilayer ceramics) on switching supplies
with some additional electrolytics if needed for bulk capacitance.  The
MLC capacitors have an extremely low ESR and can be more effective than
tantalum capacitors.  Also ceramic capacitors are more tolerant of applied
voltages and are often tested to 4X or more of rated value at the factory.

Mike

Re: [AVR-Chat] Re: proper power supply filtering

2005-05-16 by Dave VanHorn

At 01:23 PM 5/16/2005, Jim Wagner wrote:
>On the reverse diode between the input and output of the
>regulator: this is normally used to prevent reverse current
>flow through the regulator. This could occur if the input
>is suddenly forced to a low voltage while the output cap is
>still charged. The LM78xx family is not as sensitive to
>this as some and I rarely see this diode in 7805 circuits.

It's rare even to see it in the data sheets anymore, but if you look 
at the circuits, they all suffer from the same vulnerability.
The larger the cap on the output side (total of all caps) the greater 
the hazard.

The old motorola data sheets show it.

Re: [AVR-Chat] Re: proper power supply filtering

2005-05-16 by Mike Murphree

alwelch93021 said:
> 1. Be sure to read  the data sheet for the regulators.  I just
> redesigned my board and chose to use a national semi LDO which does  not
> like MLC caps.  I happen to also like MLC caps as generally they  work
> great.

Linear Tech has some issues like this with their linear regulators too...

Mike

Re: [AVR-Chat] Re: proper power supply filtering

2005-05-16 by Dave VanHorn

At 02:45 PM 5/16/2005, arhodes19044 wrote:
>--- In AVR-Chat@yahoogroups.com, Dave VanHorn <dvanhorn@d...> wrote:
>
> > Have a look at the design guides on the national semiconductor site
> > for their simple switchers.
> > You'll be looking at a "buck" topology.
>
>I saw those devices just today in the catalog.  I was wondering if
>they are any good.  If so, I can probably make good use of one of them!

You'll find they are pretty cool devices. (pun)

Move to quarantaine

This moves the raw source file on disk only. The archive index is not changed automatically, so you still need to run a manual refresh afterward.